A b s t r a c t. This study applies the MPSIAC semi-

quantitative model along with geographical information system

and remote sensing techniques to estimate sediment yield in a semi-

arid region in central Iran. Nine data layers of the model were

generated from Landsat ETM+ imagery, adapted regional maps and

field surveys. The GIS was applied to integrate the layers together

and generate the sediment yield map. The results showed a range of

sediment yield from 263.3 to 496.9 t km-2 year-1 with an average of

356.4 t km-2 year-1. However, it seems that descriptions of the

model are sometimes too broad for making reliable scoring. Never-

theless, this model is generally less data demanding and provides

an efficient way to estimate sediment yield in ungauged basins. It

was found that hills are the most sensitive land types to sediment

yield in the region.

K e y w o r d s: soil erosion, MPSIAC model, satellite data,

semi-arid region

INTRODUCTION

Soil erosion is one of the most significant forms of land

degradation (Erskine et al., 2002) and a severe eco-envi-

ronmental problem (Pan et al., 2006) in the world. There is an

increasing interest in improving water resource develop-

ment, watershed management, land use and productivity

(Daroussin and King, 2001). Erosion decreases soil quality

and crop production, declines on-site land value, and causes

off-site environmental damage. In order to protect lands

from further degradation and make mitigation measures

effective, it is essential to assess erosion hazard severity.

Sediment yield data and appropriate prediction tools are

fundamental requirements for planning and managing water

resource development schemes (Haregeweyn et al., 2005).

Recently, many process-based and empirical models are

proposed to describe and predict soil erosion by water, and

associated sediment yield. Since most process-based ero-

sion models, in general, are not well tested and require many

input parameters, the empirical prediction models continue

to play an important role in soil conservation planning.

Physically-based models and also the more complex con-

ceptual models, while having other merits, are not particu-

larly appropriate for estimating basin sediment yield since

many of these models focus only on a limited number of ero-

sion and sediment transporting processes (Merritt et al., 2003).

In contrast, empirical models are used due to their simple

structure and ease of application (Amore et al., 2004). How-

ever, these kinds of models are often based on multiple cor-

relations performed on site-specific empirical data, limiting

their application to the specific site of origin. Due to the

complexity of sediment yield models, many researchers

have tried to overcome these limitations by producing nu-

merical values to describe catchment characteristics. Often,

these models are a combination of descriptive and quan-

titative procedures that describe a drainage basin and result

in a quantitative or sometimes qualitative estimate of sedi-

ment yield in a basin. Therefore, these models can be

classified in general as semi-quantitative. The low data re-

quirements and the fact that practically all significant ero-

sion processes are considered makes semi-quantitative mo-

dels especially suited for estimating off-site effects of soil

erosion (de Vente and Poesen, 2005).

The universal soil loss equation (USLE) is the most

widely used empirical model (Wischmeier and Smith,

1965). In different parts of the world, USLE (Erskine et al.,

2002; Lee, 2004) and revised USLE (Renard, 1997) has

Int. Agrophys., 2011, 25, 241-247

Sediment yield estimation using a semi-quantitative model

and GIS-remote sensing data*

M. Mahmoodabadi

Department of Soil Science, Agriculture Faculty, Shahid Bahonar University of Kerman, Kerman, Iran

Received July 18, 2010; accepted October 4, 2010

© 2011 Institute of Agrophysics, Polish Academy of Sciences

Corresponding author’s e-mail: mahmoodabadi@mail.uk.ac.ir

*This work was partly supported by Ministry of Sciences and

Technology, Tehran, Iran, project No. 790, 2002-2005.

IIINNNTTTEEERRRNNNAAATTTIIIOOONNNAAALLL

AAAgggrrroooppphhhyyysssiiicccsss

www.international-agrophysics.org

been used by some researchers. Some other models have

been applied for estimating erosion severity and sediment

yield in sub-catchment area for which hydrometric data is

not available. In this regard, the semi-quantitative models

benefit from a more quantitative description of factors used

to characterize the basin. Probably the most well known semi-

quantitative model was developed by the Pacific South-

west Inter-Agency Committee (PSIAC, 1968) for applica-

tion in arid and semi-arid areas in the south western US (de

Vente and Poesen, 2005) and is believed to be appropriate

for the same environmental conditions in Iran (Tangestani,

2006). Unlike USLE and its different versions, the PSIAC

model can be applied to estimate total annual sediment yield

(Woida and Clark, 2001; Mahmoodabadi et al., 2005), and

not just that resulting from sheet and rill erosion. The PSIAC

model was developed for watersheds bigger than 30 km2,

however its application in smaller catchments had also good

results (Tangestani, 2006). This model is factor-based and

consists of a rating technique that characterizes a drainage

basin in terms of sensitivity to erosion, possibilities for

sediment transport and floodplain storage, the protective role

of vegeta- tion, and the influence of human land use

practices (PSIAC, 1968). The procedure considers nine factors

that depend on sur- face geology, soils, climate, runoff,

topography, ground cover, land use, channel erosion, and

upland erosion (Table 1). After summation of the individual

scores, the sediment yield rating index can be determined

and converted into an esti- mated sediment yield. A

modified version of PSIAC was presented by Johnson and

Gebhardt (1982) who produced a numerical basis for

calculating the score rating for indi- vidual factors to reduce

the subjectivity of the assessment, so called the modified

PSIAC method (MPSIAC). In Iran, some studies have been

conducted to evaluate the semi- quantitative PSIAC model

and its modified version and to test its accuracy in estimation

of sediment yields (Mahmooda- badi et al., 2005;

Tangestani, 2006). The results of these stu- dies indicate that

the model represents better estimation in comparison to

other existing models.

Any model for computing soil erosion and sediment

yield must deal with a large number of variables. The GIS

allows simpler and faster data and parameter management.

Therefore, GIS is a useful tool for sediment yield studies.

Applications of GIS and remote sensing techniques in

erosion and sediment yield assessment have been developed

recently (Lee, 2004). The combined use of remote sensing,

GIS and erosion models has been shown to be an effective

approach for estimating the severity and spatial distribution

of erosion. Clark (1999) compared the sediment yield of the

PSIAC model and the observed data in two basins of New

Mexico. The results indicated that applying the GIS, the dif-

ferences between predicted and observed values were 11-

14%, while without using the GIS these differences rose to

30%. Woida et al. (2001) developed a GIS-based appli-

cation to analyze the PSIAC factors in a distributed manner

at a 30 m resolution. It was found that this captured better the

variability of the watershed characteristics and provided

better estimates than without the use of the distributed data.

So, applying the GIS and remote sensing would improve the

prediction of sediment yield.

The purpose of this study was to assess the MPSIAC

semi-quantitative model using GIS and remote sensing for

erosion and sediment yield prediction.

MATERIALS AND METHODS

The study was performed in the Golabad watershed

which is located between 51° 56’ and 52° 14’ E and 33° 01’

and 33° 22’ N with an area of 582.7 km2

in a semi-arid region

242 M. MAHMOODABADI and A.H. CHARKHABI

MPSIAC factor Equation and description

Surface geology Y1=X1, where X1 is a geology erosion index based on rock type, hardness, fracturing and weathering

from geological reports (hard massive rock has an index of one and marine shale, mudstone or siltstone

has an index of 10)

Soils Y2=16.67X2, where X2 is the USLE soil erodibility factor values determined by procedures of Wischmier

and Smith (1978)

Climate Y3=0.2X3, where X3 is 2 year, 6 h precipitation amount in mm determined from weather records

Runoff Y4=0.2X4, where X4 is the sum of yearly runoff volume in mm times 0.03 and of yearly peak stream flow

in m3 sec-1 km-2 times 50

Topography Y5=0.33X5, where X5 is slope steepness in percent

Ground cover Y6=0.2X6, where X6 is bare ground in percent

Land use Y7= 20-0.2X7, where X7 is canopy cover in percent

Upland erosion Y8=0.25X8, where X8 is the Soil Surface Factor (SSF) determined by procedures described in Bureau

of Land Management (BLM), Manual 7317

Channel erosion Y9=16.67X9, where X9 is the SSF gully rating associated with X8

T a b l e 1. MPSIAC factor ratings (Johnson and Gebhardt, 1982)

in central Iran (Fig. 1). The elevation of this region varies

from 1 653 to 2 947 m a.s.l. and the average precipitation is

about 170 mm year-1

. Rangeland is the dominant land use in

the watershed, while due to specific climate conditions, land

use, and high grazing rate, there is poor vegetative cover in

the watershed. Based on soil taxonomy, the common soil

orders are Aridisols and Entisols. This is as a result of

unfavourable conditions of soil development such as limited

soil formation and considerable soil erosion rate. Conse-

quently, in some parts, in addition to the ochric, also the

calcic horizon can be observed at soil surface. The texture of

studied soils ranges from clay loam to sandy loam with

considerable amount of gravel (32%). The organic matter

content is between 0.25-0.43% and CaCO3 varies from 11 to

46%. These conditions lead to high erodibility of soils, since

it is influenced by soil properties such as organic matter,

particles dispersibility, and aggregate stability (Dexter and

Czy¿, 2000; Igwe, 2000; Igwe and Udegbunam, 2008).

In this study, integrated land and water irrigation system

(ILWIS) by Westen and Farifteh (1997) academic version 3,

was used to construct different digital layers and databases.

Besides, remote sensing data were derived from enhanced

thematic mapper plus (ETM+) data of the Landsat satellite.

The input data to the model were prepared by means of maps

eg topography, geology, land use, etc., reports and field

reconnaissance. Land capability and resources map (scale of

1:250 000) of the watershed was selected as the base map

(Fig. 2). As shown in this map, mountains (1-2, 1-3, 1-4),

hills (2-1, 2-2, 2-4), plateau (3-1, 3-2, 3-4) and alluvial fans

(8-1, 8-2) are four dominant land types in the region. In order

to check the boundaries of the map units, ETM+

data were

used. Different combinations of 7 bands were constructed to

obtain the best colour composite map on the basis of the least

correlation. According to correlation matrix, a colour com-

posite image created by bands of 1, 5, and 7, had the most

useful information for our purposes. After digitizing the

land capability map as a base map, it was crossed to the

colour composite map of the satellite imagery. Due to some

deviations of boundaries, the base map was adapted using

colour composite and topography maps. This modified

version map of land units is shown in Fig. 3.

To apply the MPSIAC model, the nine input layers were

created based on the descriptions given in Table 1. In order

to run the model in the GIS framework, each layer was

digitised and scored, individually. The sediment yield rating

index map was prepared by integrating these nine layers.

Annual sediment yield was calculated using the following

equation (Johnson and Gebhardt, 1982):

S eyR

= 0.253 0.036( ) , (1)

where: S y is the sediment yield (t ha-1

), assuming a sedi-

ment volume-weight of 1.360 kg m-3

, e is the base of natural

logarithms, and R (sediment yield rating index) is the sum of

MPSIAC factors.

RESULTS AND DISCUSSION

Table 2 shows nine scored-layers for different land

types and the land units. The present watershed has diverse

geology structures, while most parts of the basin are formed

of quaternary deposits with high sensitivity to erosion. By

soil sampling and field experiments, the erodibility factor in

USLE for each land unit was determined and then the soil

layer was obtained. Erodibility was assessed by the USLE K

factor. The soil-erodibility factor (K) is the rate of soil loss

SEDIMENT YIELD ESTIMATION 243

Fig. 1. Location of the Golabad watershed in Iran.

244 M. MAHMOODABADI and A.H. CHARKHABI

Fig. 2. Land capability and resources map of the Golabad water-

shed. The dominant land types are mountains: 1-2, 1-3, 1-4 land

units; hills: 2-1, 2-2, 2-4 land units; plateau: 3-1, 3-2, 3-4 land

units; and alluvial fans: 8-1, 8-2 land units.

Fig. 3. Adapted land units map of the Golabad watershed. Expla-

nation as in Fig. 1.

Land type Land unitSurface

geology Soils Climate Runoff TopographyGround

cover Land useUpland

erosion

Channel

erosion

Mountain

1-2 5.4 7.33 3.97 0.60 9.79 6.98 17.2 11.8 11.69

1-3 5.2 3.33 3.97 2.41 9.18 8.32 16.2 13.8 11.69

1-4 4.7 3.33 3.97 1.53 4.17 10.90 17.3 13.8 11.69

Hill

2-1 7.3 4.83 3.97 0.67 5.04 8.91 15.0 19.5 16.70

2-2 7.2 5.33 3.97 0.68 4.52 11.00 15.6 14.3 13.36

2-4 6.4 6.17 3.97 1.50 4.51 11.50 16.6 17.0 15.03

Plateau

3-1 8.4 5.50 3.97 0.55 0.94 10.10 15.8 10.5 10.02

3-2 8.2 3.67 3.97 0.26 1.29 11.00 15.7 11.0 10.02

3-4 6.3 3.83 3.97 2.36 3.23 13.90 18.0 13.3 13.36

Alluvial

fan

8-1 8.4 4.67 3.97 1.10 2.57 8.11 14.3 11.3 11.69

8-2 6.9 4.67 3.97 0.21 3.69 13.80 16.5 11.3 11.69

Min 4.7 3.33 3.97 0.21 0.94 6.98 14.3 10.5 10.02

Max 8.4 7.33 3.97 2.41 9.79 13.90 18.0 19.5 16.70

Average 6.8 4.79 3.97 1.08 4.45 10.40 16.2 13.4 12.45

T a b l e 2. Scores for the nine input layers of different land units

per rainfall erosion index unit plot. Division of the sub-

sequent K-factor with the factor 7.59 will yield K values ex-

pressed in SI units of t ha h ha-1

MJ-1

mm-1

(Renard et al.,

1997). Determined erodibility of the soils ranged from

0.026-0.058 in SI units indicating some differences in the

surface soil properties.

A sediment yield rating (R) layer was prepared by in-

tegrating the nine layers of the model. Table 3 shows

estimated sediment yield values for different land units. The

values of R varied from 65.07 to 82.71 with an average of

73.5. As shown in Table 3, this rating was the least for

plateaus (land units 3-2 and 3-1), while hills (2-4 and 2-1

land units) had the highest value ratings. Using Eq. (1), the

sediment yield values were calculated for different land

units. According to Table 3, estimated sediment yield

ranged from 263.3 to 496.9 with an average of 356.4 t km-1

year-1

. In spite of low annual precipitation, this range

appears to be quite high, while other studies in Iran or other

countries resulted in similar or even higher vales of sediment

yield. Erskine et al. (2002) studied the sedimentation of

dams in small sandstone drainage basins near Sydney,

Australia. Their result indicated that cultivated basins

produce an average sediment yield of 7.1 t ha-1

year-1

whereas grazed pasture and forest/wood land basins export

averages of only 3.3 and 3.1 t ha-1

year-1

, respectively.

Nevertheless, these yields were high by Australian stan-

dards. In Iran, based on 20 years data of 120 sedimentology

stations, Jalalian et al. (1994) reported a sediment yield of

348 t km-2

year-1

, with maximum value of 750 t km-2

year-1

.

In addition, the Consulting Company of Jamab (1999), using

data of 360 sedimentology stations, estimated the range of

sediment yield from 1.58 to 3 025 t km-2

year-1

. The study of

Arabkhedri et al. (2005) resulted in an average of sediment

yield of 214 t km-2

year-1

indicating an erosion rate of 6 t ha-1

year-1

. In a recent comprehensive research which has been

done by the SCWMRI (2007) and covered the whole terri-

tory of Iran, the rate of soil erosion 6.2 t ha-1

year-1

has been

reported. Our results indicated that land units 2-4 and 2-1

(hills) had the highest sediment yield. This high sediment

yield appeared to be partly due to upland erosion and chan-

nel erosion in the watershed (Table 2). In addition, vegeta-

tion cover in the region is scattered, which influences the

sediment yield. Vegetative covers have the ability to reduce

flow velocity, which reduces the erosive force of runoff (Pan

and Shangguan, 2005). In addition, the soil structure under

vegetative cover is improved, with a higher infiltration rate

and plant roots reinforcing the soil body. In general, vege-

tation reduces runoff due to canopy interception and higher

infiltration rate associated with improved soil structure (Pan

et al., 2006).

The sediment yield map based on the MPSIAC model

was generated through collapsing the total values into five

classes based on the model tables (Fig. 4). Most of the water-

shed area is classified as areas with moderate to high sedi-

ment yields. Considering the fact that the soil formation rate

SEDIMENT YIELD ESTIMATION 245

Land typeLand

unit

Area

(km2)

Sediment

yield (R)

Sediment yield

(km-2 year-1)

Mountain

1-2 36.4 74.68 372.1

1-3 164.5 74.05 363.8

1-4 41.4 71.32 329.7

Hill

2-1 47.4 81.90 482.6

2-2 28.6 75.96 389.7

2-4 65.5 82.73 497.2

Plateau

3-1 57.9 65.83 270.6

3-2 68.0 65.04 263.0

3-4 20.3 78.23 422.9

Alluvial

fan

8-1 48.8 66.02 272.5

8-2 3.9 72.67 346.2

Min - 65.07 263.3

Max - 82.71 496.9

Average - 73.48 356.4

T a b l e 3. Sediment yield results of the MPSIAC model

application

Fig. 4. Sediment yield map of the Golabad watershed: VL – very

low, L – low, M – moderate, H – high, and VH – very high.

is very limited in this environment, obviously the high

values of sediment yields could be problematic for a sustaina-

ble development system in this watershed. Therefore, some

conservation and management practices are needed to be

used in the hilly land units which have high sediment yields.

Usually, soil erosion and sediment yields cannot be

estimated with high-levels of confidence, using the existing

grey-box models, models with semi-quantitative estimating

capabilities such as MPSIAC. Basin sediment yield is a pro-

duct of all sediment producing and transport processes

within a basin, which are rather difficult to model in arid

environments due to intermittent stream flow, the discon-

tinuity of flow, and highly irregular rainfall distributions (de

Vente and Poesen, 2005). This leads to a necessity for cali-

bration of a model outside the area for which the model is

developed. Moreover, involving experienced and related

experts during the rating of individual scores can minimize

subjectivity of the scoring processes.

Until now it has not been possible to develop a model

consisting of all components together, that could be used at

the basin scale with reasonable results. The difficulties are

due to a combination of natural complexity, spatial hetero-

geneity, and lack of available data (de Vente and Poesen,

2005). The PSIAC model focuses on sediment yield at a ba-

sin scale and off-site effects, as it does not specifically iden-

tify spatially distributed sources of sediments. In addition,

the model specifically considers contribution of landslides,

though only through observation of their occurrences. Thus,

this model can be considered as a holistic sediment yield

model, trying to include most erosion and sediment trans-

port processes. It seems that descriptions of the PSIAC are

sometimes too broad, which makes scoring difficult, and it

results in poorer model performance. The model also has

some extrapolation problems, especially when it relates to

a score for measuring sediment yield. Nevertheless, this

model is generally less data-demanding and it is an effort to

represent available knowledge on sources of sediment and

sediment transport, which certainly could be a useful tool

supporting decision making processes in a sustainable

watershed management system.

CONCLUSIONS

1.The estimations of the MPSIAC model showed that

the sediment yield varied from 263.3 to 496.9 with an ave-

rage of 356.4 t km-2

year-1

.

2. The results indicated that hills have the highest sensi-

tivity to erosion and sediment yield.

3. Preliminary results of the applicability of this model

showed it can be recommended for arid and semiarid water-

sheds with ungauged basins. However, there are still some

uncertainties in the structure of the model, which can be only

identified by retesting the model using the same procedure

employed in its development under different environmental

conditions.

4. Moreover, application of GIS and remote sensing

indicated that these are helpful techniques to estimate sedi-

ment yields for better soil conservation practices in water-

sheds such as this. Hence, the combined use of remote sensing

and GIS could be an effective approach for estimating the

severity of erosion in ungauged watersheds of central Iran.

REFERENCES

Amore E., Modica C., Nearing M.A., and Santoro V.C., 2004.

Scale effect in USLE and WEPP application for soil erosion

computation from three Sicilian basins. J. Hydrol., 293,

100-114.

Arabkhedri M., Iranmanesh F., Razmju P., and Hakimkhani S.,

2005. Preparation of sediment yield map of Iran and

watersheds prioritizing in sediment yield. 3rd Nat. Cong.

Erosion and Sediment, August 28-31, Tehran, Iran.

Clark K.B., 1999. An estimate of sediment yield for two small

watersheds in a geographic information system. MSc. Thesis,

Geography University of New Mexico, Mexico.

Consulting Company of Jamab., 1999. Comprehensive Project of

Water, Tehran, Iran.

Daroussin J. and King D., 2001. Mapping erosion risk for culti-

vated soil in France. Catena, 46, 207-220.

de Vente J. and Poesen J., 2005. Predicting soil erosion and sedi-

ment yield at the basin scale: Scale issues and semi-

quantitative models. Earth Sci. Rev., 71, 95-125.

Dexter A.R. and Czy¿ E.A., 2000. Effects of soil management on

the dispersibility of clay in a sandy soil. Int. Agrophysics, 14,

269-272.

Erskine W.D., Mahmoudzadeh A., and Myers C., 2002. Land

use effects on sediment yields and soil loss rates in small

basins of Triassic sandstone near Sydney, NSW, Australia.

Catena, 49, 271-287.

Haregeweyn N., Poesen J., Nyssen J., Verstraeten G., de Vente J.,

Govers G., Deckers S., and Moeyersons J., 2005. Specific

sediment yield in Tigray-Northern Ethiopia: Assessment

and semi-quantitative modeling. Geomorphology, 69,

315-331.

Igwe C.A., 2000. Runoff, sediment yield and erodibility characte-

ristics of some soils of central eastern Nigeria under simula-

ted rainfall. Int. Agrophysics, 14, 57-65.

Igwe C.A. and Udegbunam O.N., 2008. Soil properties influen-

cing water-dispersible clay and silt in an Ultisol in southern

Nigeria. Int. Agrophysics, 22, 319-325.

Jalalian A., Ghahareh A.M., and Karimzadeh H., 1994. Erosion

and sediment yield and reasons in the watersheds of Iran.

Proc. 4th Iranian Soil Cong. August 28-31, Isfahan, Iran.

Johnson C.W. and Gebhardt K.A., 1982. Predicting Sediment

yields from sagebrush rangelands. Proc. Workshop Estima-

ting Erosion and Sediment Yield on Rangelands, March

12-14, Tucson, AZ, USA.

Lee S., 2004. Soil erosion assessment and its verification using the

universal soil loss equation and geographic information system:

a case study at Boun, Korea. Environ. Geol., 45, 457-465.

Mahmoodabadi M., Charkhabi A.H., Rafahi H.G., and Gorji M.,

2005. Zonation of soil erosion hazard in Golabad watershed

using MPSIAC model and geographical information system.

J. Agric. Sci. Iran, 36, 511-520.

246 M. MAHMOODABADI and A.H. CHARKHABI

Merritt W.S., Letcher R.A., and Jakeman A.J., 2003. A review

of erosion and sediment transport models. Environ. Model.

Software, 18, 761-799.PSIAC (Pacific Southwest Inter-Agency Committee), 1968. Fac-

tors affecting sediment yield in the pacific southwest areaand selection and evaluation of measures for reduction oferosion and sediment yield. Report of the Water Mana-gement Subcommittee, PSIAC Press, Boise, ID, USA.

Pan C., Shangguan Z., and Lei T., 2006. Influences of grass and

moss on runoff and sediment yield on sloped loess surfaces

under simulated rainfall. Hydrol. Proc., 20, 3815-3824.

Pan C.Z. and Shangguan Z.P., 2005. Influence of forage grass on

hydrodynamic characteristics of slope erosion. J. Hydraulic

Eng., 36, 271-277.

Renard K.J., Foster G.A., Weesies G.A., and McCool D.K.,

1997. Predicting Soil Erosion by Water: a Guide to Conser-

vation Planning with Revised Universal Soil Loss Equation.

USDA Agriculture Handbook, Washington, DC, USA.

SCWMRI, 2007. Aspects of Iranian Watersheds. Final Project of

Erosion and Sediment. Forests, Rangelands and Watershed

Management Organization Press, Tehran, Iran.

Tangestani M.H., 2006. Comparison of EPM and PSIAC models

in GIS for erosion and sediment yield assessment in a semi-

arid environment: Afzar Catchment, Fars Province, Iran.

J. Asian Earth Sci., 27, 585-597.

Wischmeier W.H. and Smith D.D., 1965. Predicting rainfall-

erosion losses from cropland east of the Rocky Mountains:

guide for selection of practices for soil and water conser-

vation, US Dept. Agricultural Handbook, Washington, DC,

USA.

Woida K. and Clark, K.B., 2001. GIS adaptation of PSIAC for

predicting sediment yield rates in New Mexico. Proc. 7th

Conf. Federal Interagency Sedimentation , May 1-3, Reno,

NV, USA.

SEDIMENT YIELD ESTIMATION 247